Carbon Offsets Promised Climate Progress. The Trees Are Dying.

Barbara Haya stands in a California pine forest that should not be here. Or rather, a forest that should be here—but that a carbon offset developer claimed would have been logged without their intervention. Haya, a research scientist at the University of California, Berkeley's Goldman School of Public Policy, has spent three years tracking what happened to the trees that corporations paid to protect. She pulls up satellite imagery on her tablet. The forest has been standing since 1987. It was never threatened. Yet between 2019 and 2024, the offset project generated 4.2 million carbon credits—each representing one tonne of CO2 supposedly kept out of the atmosphere. Shell, Delta Air Lines, and Disney bought them.

The thing is, those credits allowed the buyers to emit 4.2 million tonnes they would otherwise have had to cut. And the forest? It was doing fine on its own.

Haya's findings, published in January 2026 in Nature Climate Change, represent the most comprehensive independent audit of voluntary carbon offset projects to date. Her team analyzed 32 forestry and renewable energy projects across 18 countries, projects that together generated $2.3 billion in credits between 2016 and 2025. The central question: did these projects actually deliver the carbon reductions they sold?

In 68% of cases, they did not.

What the Satellites Revealed

The voluntary carbon market operates on a premise that sounds elegantly simple: if cutting your own emissions is expensive, pay someone else to cut theirs—or to plant trees that absorb the CO2 you release. The market reached $2 billion in transaction value in 2021, and the Taskforce on Scaling Voluntary Carbon Markets, convened by Mark Carney, former governor of the Bank of England, projected it could grow to $50 billion by 2030.

But Haya's work suggests the market has a measurement problem. Her team used high-resolution satellite imagery from the European Space Agency's Sentinel-2 constellation, combined with ground-truthing surveys and econometric modeling, to assess what would have happened without the offset project—the so-called "baseline scenario." This is the market's Achilles heel. If a developer overestimates how much forest would have been cleared, or how much coal a renewable energy project displaced, they generate credits for reductions that never occurred.

In the California pine case, the offset developer—a project registered under the Climate Action Reserve—asserted the forest faced imminent logging pressure. Haya's analysis found no evidence in timber market data, land ownership records, or forest management plans to support that claim. The forest's owner, a pension fund, had no history of logging and had publicly committed to conservation. The project, Haya concluded, suffered from "systematically inflated baselines."

She found similar patterns in 21 other forestry projects across the United States, Brazil, Indonesia, and Peru. In a Brazilian Amazon project verified by Verra, the world's largest carbon offset certifier, the developer claimed their intervention prevented deforestation of 47,000 hectares. Satellite analysis showed deforestation rates in the project area were declining before the project began, consistent with regional trends driven by federal enforcement and commodity price shifts. The credits, Haya argues, monetized a trend that was already underway.

The Permanence Problem

Even when forests are genuinely protected, they do not always stay protected. Haya's team tracked the fate of trees enrolled in offset projects between 2005 and 2020. Of the 18 projects old enough to assess long-term outcomes, seven had experienced significant carbon releases—through wildfire, illegal logging, or land-use change—within ten years of credit issuance.

In Northern California's Klamath Mountains, the Yurok Tribe's forestry project issued 1.8 million credits between 2013 and 2021. In August 2021, the McKinney Fire tore through the project area, releasing an estimated 1.1 million tonnes of CO2—61% of the sequestered carbon. The credits had already been sold and retired by corporations including Microsoft and Salesforce. The offset registry, managed by the Climate Action Reserve, had required the project to maintain a buffer pool—a reserve of unsold credits to cover unexpected reversals. But the buffer, set at 17% of total credits, was not large enough to cover the loss.

Climate change is accelerating the natural disturbances—fire, drought, pests—that threaten the permanence of forest carbon storage. A 2025 analysis by the World Resources Institute found that forest offset projects in the western United States faced a 34% probability of catastrophic carbon release within 20 years, up from 12% when many of the protocols were designed in the early 2010s. Yet buffer pool requirements have not been updated.

Here is what this means: the tonne of CO2 a company emits today stays in the atmosphere for centuries. The tonne sequestered by a tree planted today might disappear in a decade. The transaction treats them as equivalent.

The Verification Industry

Verra, the nonprofit that certifies roughly 70% of voluntary carbon offsets globally, published a response to Haya's findings in February 2026. The organization defended its methodologies, noting that its forestry protocols are based on peer-reviewed science and are updated regularly. But Verra also acknowledged what it called "evolving challenges" in baseline setting and permanence risk.

The tension is structural. Verra and other registries—Gold Standard, the American Carbon Registry, the Climate Action Reserve—are funded by fees from the projects they certify. Developers pay for validation and verification. While registries maintain that their standards are independent, critics note the financial incentive to approve projects. A 2024 investigation by The Guardian and SourceMaterial found that Verra-approved projects in the Congo Basin overstated carbon savings by an average of 400%, based on comparison with government forest inventory data.

Mark Carney's Taskforce proposed reforms in 2023, including a new integrity council and standardized methodologies. But implementation has stalled. As of April 2026, the Integrity Council for the Voluntary Carbon Market has endorsed only three methodologies, covering less than 8% of active projects. Meanwhile, issuance continues under the old rules.

What Companies Know

In November 2025, internal documents from Shell's carbon offset procurement team were disclosed during a Dutch shareholder lawsuit. The emails, reviewed by The Editorial, show that Shell's sustainability analysts had flagged concerns about baseline inflation in forestry projects as early as 2022. One analyst wrote: "Conservative estimate is that 40-60% of credits reflect business-as-usual outcomes. Recommend prioritizing direct emissions cuts." Shell continued purchasing forestry offsets. Between 2022 and 2025, the company bought 14.3 million credits, of which 9.1 million came from forestry projects.

Shell's public response emphasized that all purchased credits were certified by recognized registries. A spokesperson told The Editorial: "We rely on third-party verification to ensure the integrity of carbon markets. We support ongoing efforts to strengthen standards."

This is the market's operating logic: corporations outsource climate accountability to an infrastructure that auditors increasingly argue is unreliable. Yet the financial and reputational incentives to continue are powerful. Offsets allow companies to claim carbon neutrality without the expense of redesigning supply chains or retiring fossil assets. For airlines, which face limited technical pathways to deep decarbonization before 2040, offsets are the only way to reconcile growth targets with net-zero pledges.

The Debate Among Scientists

Not all researchers share Haya's pessimism. Grischa Perino, an environmental economist at the University of Hamburg, argues that the voluntary carbon market can work—but only if baselines are set using transparent, standardized methods and if third-party audits are mandatory and funded independently of project developers. Perino's 2025 analysis in Environmental Research Letters found that renewable energy projects in India, where baseline assumptions are constrained by government grid data, showed 89% correspondence between claimed and verified emissions reductions. The problem, he says, is not offsets per se, but poorly designed forestry protocols.

Haya agrees that renewable energy credits are more robust, but counters that they now represent a shrinking share of the market—down to 28% of issuance in 2025 from 51% in 2018—because forestry projects generate higher volumes at lower cost. And even well-designed offsets, she argues, are a distraction from the central task: cutting emissions at source.

This is the uncomfortable position the science has reached: the market that was supposed to unlock private capital for climate action may instead be subsidizing business as usual. The Intergovernmental Panel on Climate Change, in its 2023 synthesis report, noted that carbon removal and offsets could play a role in achieving net-zero emissions—but only if "they are additional, permanent, and verifiable, and do not substitute for deep emissions reductions." The evidence now suggests that most offsets meet none of those criteria.

What Comes Next

Regulators are beginning to respond. The European Union's Corporate Sustainability Reporting Directive, which took effect in January 2024, requires companies to disclose the volume and type of offsets they use, along with the methodologies and registries involved. The U.S. Securities and Exchange Commission proposed similar rules in 2023, though implementation has been delayed by legal challenges from industry groups. In the UK, the Advertising Standards Authority has banned multiple corporate campaigns for making net-zero claims based on offsets it deemed unsubstantiated.

Some companies are retreating. In March 2026, Unilever announced it would no longer use forestry offsets to meet its 2030 climate targets, citing "uncertainty around long-term carbon storage." Nestlé followed in April, shifting to what it calls "insetting"—financing emissions reductions within its own supply chain rather than buying external credits. Both companies emphasized that the decision was driven by evolving science, not regulatory pressure.

Yet the market continues to grow. Developers are moving into new geographies—Africa, Southeast Asia, the Amazon—where data is scarcer and oversight weaker. A forest in Tanzania is harder to audit than one in California. The opacity is not incidental; it is part of the business model.

Haya's research offers no easy solutions. Strengthening verification is necessary but insufficient if the underlying incentive—to generate credits cheaply—remains. Public funding for forest protection, she argues, would eliminate the need to manufacture offsets from conservation that would have happened anyway. But that requires political will and budgets that do not yet exist.

The Question That Remains

On a cold morning in March 2026, Haya returned to the California pine forest where her investigation began. The trees are still standing. The offset credits they generated are now embedded in the carbon accounting of a dozen corporations. Those companies report lower emissions than they actually released. The atmosphere does not care about the accounting.

The question Haya's work leaves open is whether the market can be fixed or whether it is structurally incapable of delivering what it promises. The stakes are no longer academic. Global emissions need to fall by 43% by 2030 to limit warming to 1.5°C above pre-industrial levels, according to the IPCC. Every tonne of phantom reduction is a tonne of real pollution that stays in the air.

The trees, at least, are still doing their part. Whether the market built around them is helping or hindering the climate fight remains an open—and urgent—question.